Fix input validation to support passing device_mesh to FSDP
#19392
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
awaelchli
added
strategy: fsdp
github-actions
bot
added
fabric
awaelchli
added
feature
github-actions
bot
added
fabric
awaelchli
added
bug
awaelchli
changed the title
device_mesh to FSDPdevice_mesh to FSDP
awaelchli
requested review from
carmocca,
justusschock,
Borda and
williamFalcon
as code owners
Codecov Report
Additional details and impacted files@@ Coverage Diff @@
## master #19392 +/- ##
==========================================
- Coverage 84% 54% -30%
==========================================
Files 448 443 -5
Lines 37910 37817 -93
==========================================
- Hits 31659 20329 -11330
- Misses 6251 17488 +11237 |
carmocca
approved these changes
Feb 2, 2024
Co-authored-by: Carlos Mocholí <[email protected]>
Borda
approved these changes
Feb 2, 2024
7 tasks
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
What does this PR do?
When you set
FSDPStrategy(sharding_strategy=HYBRID_SHARD)and neither pass a autowrap policy nor a process group tuple, Lightning raises a RuntimeError. But now that PyTorch 2.2 supports device_mesh, we should be allowed to pass a device mesh instead of a process group. This PR enables that.This is technically a bugfix because device_mesh was already supported experimentally before PyTorch 2.2.
📚 Documentation preview 📚: https://pytorch-lightning--19392.org.readthedocs.build/en/19392/
cc @Borda @carmocca @justusschock @awaelchli